Iron(iii) citrate speciation in aqueous solution

PAPER www.rsc.org/dalton | Dalton Transactions

Iron(III) citrate speciation in aqueous solution

Andre M. N. Silva, XiaoLe Kong, Mark C. Parkin, Richard Cammack and Robert C. Hider*

Received 3rd June 2009, Accepted 21st July 2009First published as an Advance Article on the web 19th August 2009DOI: 10.1039/b910970f

Citrate is an iron chelator and it has been shown to be the major iron ligand in the xylem sap of plants.Furthermore, citrate has been demonstrated to be an important ligand for the non-transferrin boundiron (NTBI) pool occurring in the plasma of individuals suffering from iron-overload. However, ferriccitrate chemistry is complicated and a definitive description of its aqueous speciation at neutral pHremains elusive. X-Ray crystallography data indicates that the alcohol function of citrate (Cit4-) isinvolved in Fe(III) coordination and that deprotonation of this functional group occurs upon complexformation. The inability to include this deprotonation in the affinity constant calculations has been amajor source of divergence between various reports of iron(III)–citrate affinity constants. However therecent determination of the alcoholic pKa of citric acid (H4Cit) renders the reassessment of the ferriccitrate system possible. The aqueous speciation of ferric citrate has been investigated by massspectrometry and EPR spectroscopy. It was observed that the most relevant species are a monoirondicitrate species and dinuclear and trinuclear oligomeric complexes, the relative concentration of whichdepends on the solution pH value and the iron : citric acid molar ratio. Spectrophotometric titrationwas utilized for affinity constant determination and the formation constant for the biologically relevant[Fe(Cit)2]5- is reported for the first time.

Introduction

Citric acid is a ubiquitous molecule of biological relevance. Centralto one of the biological roles of citrate is its ability to chelate metalsand in particular iron. Citrate binds to Fe(II) at the active site ofaconitase1 and has been proposed to be a constituent of the lowmolecular weight cytosolic iron pool.2

The ability of citrate to solubilize ferric hydroxide has beenexploited by both plants and microbes. Citric acid is found inthe root exudates of many plants3 and the mobilized iron istransported in the xylem sap as ferric citrate.4 In bacteria, citricacid acts as an exogenous siderophore. Escherichia coli possessesa specific transport system for ferric citrate5 and Bradyrhizobiumjaponicum was found to release citric acid under iron deficientgrowth conditions.6 This unique ability to bind ferric iron seemsto have led to the evolution of a class of citrate related siderophores,such as rhizoferrin and staphyloferrin A, where the structure ofthe chelating unit preserves the citric acid backbone.7 In humans,Fe(III) binding by citrate is of particular significance under iron-overload conditions. Citrate is present in the blood plasma at0.1 mM and it has been shown to be the major low molecularweight ligand for the non-transferrin bound iron (NTBI) pool inthe plasma of iron-overloaded patients.8

Despite its central role in iron metabolism the aqueousspeciation of ferric citrate remains unclear. The chemistry ofFe(III) complexation by citrate is complicated9 and disagreementprevails in the literature, especially for the physiological pHvalues. Most of the early studies were centred on potentiometrictitrations of Fe(III)–citrate. In 1948, Lanford and Quinan reported

Pharmaceutical Sciences Research Division, King’s College London,Franklin-Wilkins Building, 150 Stamford Street, London, UK SE1 9NHE-mail: [email protected]; Fax: +44 2078484800; Tel: +442078484882

the existence of a complex where iron and citrate existed in amolar ratio of 1 : 1.10 This was later confirmed by other authorswho have found [Fe(Cit)]0, [Fe(HCit)]+ and [Fe(Cit)(OH)]-, whereH3Cit is the form for citric acid, claimed to be the relevantspecies.11,12 Controversy occurs, however, concerning the extentof hydrolysis13 occurring in the complex, with Timberlake findinga 2 : 2 complex.14 Martin has argued that the dominant speciesunder neutral conditions is similar to that at lower pH values (1 : 1complex),15 but more recently the existence of a mononucleardicitrate16 and a dinuclear complex have been reported.17

Most of the previously described studies were carried out insolutions where the Fe(III) : citrate molar ratio was higher than1 : 6 under low pH conditions. Spiro et al.18 have reported that insolutions where Fe(III) and citrate are present, at close to equimolarconcentrations, increasing pH promotes the formation of largeferric citrate polymers resulting from extensive iron hydrolysis.However, on increasing the excess of citrate, the formation of a[Fe(Cit)2]5- complex, where the loss of citrate hydroxyl protonwas observed, competed with polymer formation. When theconcentration of citrate exceeded that of Fe(III) by more than 20fold, no polymer could be observed.19

To date, the structures of four distinct ferric citrate com-plexes have been obtained by X-ray crystallography. Underacidic conditions two dinuclear complexes, [Fe2(Cit)2(H2O)2]2- and[Fe2(HCit)3]3-,20 and one nonairon complex, [Fe9O(Cit)8(H2O)3]7-

have been prepared.21 The mononuclear dicitrate [Fe(Cit)2]5-

complex was crystallized at pH values closer to neutral.22 All fourstructures clearly show that citrate chelates Fe(III) with loss ofthe alcoholic proton. More recently, Gautier-Luneau et al. havefurther clarified the aqueous speciation of ferric citrate by sys-tematizing the procedures for the crystallization conditions of thedifferent complexes and by analyzing the speciation of such speciesutilizing mass spectrometry.23 Mass spectrometry indicates that the

8616 | Dalton Trans., 2009, 8616–8625 This journal is © The Royal Society of Chemistry 2009

[Fe(Cit)2]5- complex is likely to be an important species under phy-siological conditions. Complications associated with the alcoholdeprotonation of citric acid have prevented determination of thestability constant of this species. This problem may also accountfor the discrepancies observed in the different speciation studies.

In the present study, we report on the determination of thestability constant for the mononuclear dicitrate complex andpropose a speciation model that appears to be in agreementwith all the available evidence. Electrospray ionisation-massspectrometry (ESI-MS) studies were carried out on an ion-trapmass spectrometer, confirming the results by Gautier-Luneauet al.23 and the resulting information was utilized to interpretthe data resulting from spectrophotometric titration. Electronparamagnetic resonance spectroscopy (EPR) demonstrated theexistence of a mononuclear iron complex at pH = 7.4, reinforcingits relevance for the understanding of iron citrate chemistry inbiological systems.

Experimental

Materials

All reagents were purchased from Aldrich Chemicals with thehighest degree of purity and lowest iron contamination. Methanol(HPLC grade) was obtained from Fisher Scientific and aqueousammonia solution (30%, AnalaR) was acquired from BDH. Fortitrations, standard potassium hydroxide aqueous solution with anominal concentration of 10 M was used (Fisher Scientific).

Electrospray ionisation-mass spectrometry of ferric citrate aqueoussolutions

ESI-MS experiments were performed on a Thermo Fisher Sci-entific Deca-XP ion trap mass spectrometer equipped with anelectrospray source. Mass spectra of the ferric citrate sampleswere obtained in the range 50–1300 m/z. Samples were infuseddirectly into the ESI source utilizing the syringe pump on themass spectrometer at 10 mL min-1. Solutions were analyzed inthe negative ionization mode. The temperature of the heated iontransfer capillary and the electrospray voltage were optimized foreach individual sample as these were dependant on the conditions(iron : citrate molar ratio and pH value) of the solution beinganalyzed. This meant, depending on the sample, that the capillarytemperature was between 100 to 120 ◦C, while the electrosprayvoltage was varied within the range -3.0 to -4.2 kV. Experimentalmass values throughout this work refer to the monoisotopic mass.

A series of solutions was prepared with iron–citrate molar ratiosranging from 1 : 1 to 1 : 100 and pH values varying from 4.2 to9.0. Solid Fe(ClO4)3·9H2O was dissolved in the desired volumeof a 200 mM citric acid stock solution, such as to form the finalconcentration of 1 mM iron. The solution pH value was adjustedby the addition of aqueous ammonia and the final solution volumewas made up to 25 mL by the addition of water. Solutions wereprepared 24 h before the analysis and kept in the dark to avoidiron photoreduction. Prior to infusion into the mass spectrometer,methanol was added up to 20% and the pH value was re-checked.

Electron paramagnetic resonance (EPR) studies on ferric citratesolutions

Spectra were recorded on a Bruker ESP300 spectrometer witha Bruker high sensitivity cavity (SHQEW0401) and an Oxford

Instruments ESR900 liquid helium flow cryostat at 12.5 K usingthe following running conditions: microwave power 20 mW,frequency 9.40 GHz, modulation amplitude 0.5 mT, modulationfrequency 100 kHz.

Iron–citrate solutions, with Fe : Cit molar ratio ranging from1 : 5 to 1 : 1000, were prepared by adding a suitable volume of acitric acid stock solution to an 18.068 mM iron chloride standardsolution. MOPS buffer pH 7.4 was added to the final concentrationof 100 mM and the volume was completed to 10 ml with theaddition of water. Two series of solutions were prepared containinga final iron concentration of 10 mM and 100 mM. A solution with100 mM citric acid and 100 mM MOPS buffer pH 7.4 was preparedand analyzed as a negative control.

Spectrophotometric titration of ferric citrate solutions

The automatic titration system used in this study comprises anautoburette (Metrohm Dosimat 765 l mL syringe) and MettlerToledo MP230 pH meter with Metrohm pH electrode (6.0133.100)and a reference electrode (6.0733.100). 0.1 M KCl electrolytesolution was used to maintain the ionic strength. The temperatureof the test solutions was maintained in a thermostatic jacketedtitration vessel at 25 ± 0.1 ◦C using a Techne TE-8J temperaturecontroller. The solution under investigation was stirred vigorouslyduring the experiment. A Gilson Mini-plus#3 pump with speedcapability (20 mL min-1) was used to circulate the test solutionthrough a Hellem quartz flow cuvette. For stability constantdeterminations a 50 mm pathlength cuvette was utilized. The flowcuvette was mounted on a HP 8453 UV-visible spectrophotometer.All the instruments were interfaced to a computer and controlledby a Visual Basic program. Automatic titration and spectralscans adopted the following strategy: the pH of a solutionwas increased by 0.1 pH unit by the addition of KOH fromthe autoburette, when pH readings varied by <0.001 pH unitover a three second period, an incubation period was activated;a period of five minutes was adopted. The stability of thesolutions was monitored by visible spectrophotometry and theseexperiments demonstrated that five minutes was the minimumperiod required in order to achieve equilibrium, particularly atneutral pH values. Iron hydrolysis proceeds slowly and is expectedto be minimal during this time period, a fact reinforced by thesolution stability experiments. At the end of the equilibriumperiod, the spectrum of the solution was recorded. The cycle wasrepeated automatically until the defined end point pH value wasachieved.

The iron–citrate affinity determination consisted of three titra-tions in order to determine the stability constants of MLH, M2L2

or M3L3 and ML2 respectively. In the first and second experimentsthe iron : citrate molar ratio was 1 : 20, while under the third con-dition, a molar ratio of 1 : 200 was adopted. To determine MLH,a 12.65 mL solution containing [Fe3+] = 460 mM and [Citrate] =8.84 mM was titrated with 1 M HCl from pH 2 to 0.65 yielding12 spectra. For the determination of the iron(III) affinity con-stants of multinuclear complexes, the following conditions wereadopted: [Fe3+] = 475 mM and [Citrate] = 9.12 mM (14.21 mL)were titrated with KOH (0.12 M) from pH 1.86 to 3.09, yielding10 spectra. To determine the iron(III) affinity constants of mononu-clear complex ML2, a solution containing [Fe3+] = 435 mM and[Citrate] = 87.7 mM (15.02 mL) was titrated with KOH (0.12 M)

This journal is © The Royal Society of Chemistry 2009 Dalton Trans., 2009, 8616–8625 | 8617

from pH 2.96 to 7.86, yielding 31 spectra. All the experiments werecarried out using the auto spectrophotometric titration system instandard mode.

Determination of iron(III) citric acid affinity constants

Spectrophotometric titration data was processed utilizing the soft-ware pHab as described in detail elsewhere.24 Two stoichiometricmodels were adopted, one involving the complexes ML, M2L2 andML2 while the other considered ML, M3L3 and ML2. Citrate wastreated has a tetrabasic ligand (H4L) with pKa values 3.13, 4.76,6.40 and 14.4.25,26

Speciation plots were produced with the aid of the hyperquadsimulation and speciation computer software (HySS)27 utilizingthe affinity constants previously determined for the different ferriccitrate complexes and the stability constants for the formationof iron hydroxide species, with the following logarithmic values:-2.563 [FeOH]2+, -6.205 [Fe(OH)2]+, -15.1 Fe(OH)3, -21.883[Fe(OH)4]-, -2.843 [Fe2(OH)2]4+ and -6.054 [Fe3(OH)4]5+.28

Results

Electrospray ionisation-mass spectrometry of ferric citrate aqueoussolutions

ESI-MS has previously been utilized for the determination ofthe speciation of aqueous solutions of ferric citrate.23 In thispresent study, similar conditions to those previously reportedwere adopted for the mass spectrometry and sample preparation.Solutions were prepared by dissolving a Fe(III) salt in a citric acidsolution. The pH value of the solution was carefully adjustedutilizing aqueous ammonia. Employing ammonia as the basesubstitute for the direct addition of hydroxide led to minimaliron hydrolysis. It was observed that pH adjustment with sodiumhydroxide led to the appearance of a yellow/brown colour whichwas assumed to result from the formation of ferric hydroxidespecies. Adjustment of the solution pH value utilizing aqueousammonia resulted in green solutions, as expected for ferric citratecomplexes.

Due to the low sensitivity obtained when operating the massspectrometer in the negative ion mode, solutions were prepared toa final iron concentration of 1 mM and methanol was added toa concentration of 20% before analysis. Solutions containing aniron : citrate molar ratio of 1 : 10 at different pH values were alsoanalyzed without the addition of methanol. Despite the decrease insignal to noise ratio which posed difficulties for the identification ofthe less abundant species, results were equivalent to those obtainedwith 20% methanol–water solutions, indicating that the additionof methanol did not introduce changes to the ferric citrate complexspeciation.

The different ferric citrate species identified are summarized inTable 1. Citric acid was considered to be a tetrabasic acid andis represented as H4Cit, while (Cit)4- represents the completelyionized citrate ion with a molecular form of C6H4O7

4-. Thecalculated mass to charge ratio (m/z) for the most abundantisotope is reported for each identified species while experimentalvalues are shown in the corresponding figures. Peak assignmentwas based both on the m/z value and the isotopic distributionpattern for the corresponding species.

The mass spectra obtained for solutions with a 1 : 10Fe(III) : citrate molar ratio at acidic (~4.4), neutral (~7.4) and basic(~9.0) pH values are presented in Fig. 1. The dominant peakfor all the solutions had a m/z value of 191, resulting from amonodeprotonated citric acid molecule. Due to this finding andin an attempt to provide a comprehensive view of the spectrathe m/z range 200–1000 is presented for the complexes. However,the dominant feature is still associated with a citric acid adduct(H4Cit)(H3Cit)- at m/z 383. At pH 4.3 (Fig. 1(A)) there waspredominance of dinuclear and trinuclear species with the complex[Fe2(Cit)2]2- appearing as a citric acid adduct at m/z 340 and inits protonated form [Fe2(Cit)2H]- with one, two or three citricacid molecules, respectively at m/z 681, 873 and 1064. Peaks

Fig. 1 Electrospray ionization-mass spectra of solutions with iron : citratemolar ratio 1 : 10 at different pH values. (A) pH 4.3; (B) pH 7.3; (C) pH 9.0.A relationship between low pH and an increasing number of differentspecies in solution becomes clear from the spectral analysis. Spectra arenormalized for the most abundant peak (m/z = 383). Due to the excess ofcitric acid, the scale of peak abundance is not shown to 100%, in order tofacilitate the visualization of the less abundant species.

8618 | Dalton Trans., 2009, 8616–8625 This journal is © The Royal Society of Chemistry 2009

Table 1 Ferric citrate species in aqueous solution detected by ESI-MS as a function of iron citrate molar ratio and pH. The species indicated in boldcorresponded to the ferric citrate species found to predominate in the respective mass spectrum. Values are for the calculated m/z value of each species

Fe : Cit ratio Acid pH species, m/z Neutral pH species, m/z Basic pH species, m/z

1 : 1 pH 4.5[Fe(Cit)2H3]2- 217.5[Fe3(Cit)3H]2- 366.5(H4Cit)2[Fe3(Cit)3H]2- 558.5[Fe3O(Cit)3H3]2- 375.5[Fe3O(Cit)3H4]- 752[Fe3(Cit)4H5]2- 462.5[Fe3(Cit)4H6]- 926(H4Cit)[Fe4(Cit)4H]3- 390(H4Cit)2[Fe4O(Cit)4]6- 229(H4Cit)n[Fe5O(Cit)4H]2- 620.5, 716.5

1 : 2 pH 4.2 pH 7.4 pH 9.0[Fe(Cit)2H3]2- 217.5 [Fe(Cit)2H3]2- 217.5 [Fe(Cit)2H3]2- 217.5[Fe(Cit)2H4]- 436 [Fe3(Cit)3H]2- 366.5 [Fe(Cit)2H4]- 436(H4Cit)[Fe(Cit)2H4]- 628 (H4Cit)n[Fe3(Cit)3H]2- 558.5, 654.5 (H4Cit)n[Fe(Cit)2H4]- 628, 820(H4Cit)[Fe2(Cit)2]2- 340 [Fe3O(Cit)3H3]2- 375.5 (H4Cit)3[Fe2(Cit)2H]- 1064[Fe2(Cit)2H]- 489 (H4Cit)n[Fe3O(Cit)3H3]2- 663.5, 759.5 [Fe3(Cit)3H]2- 366.5(H4Cit)n[Fe2(Cit)2H]- 873, 1064 (H4Cit)4[Fe3O(Cit)3H3]2-·H2O 768.5 (H4Cit)2[Fe3(Cit)3H]2- 558.5, 654.5, 750.5[Fe2(Cit)3H5]- 681 [Fe3(Cit)4H5]2- 462.5 [Fe3O(Cit)3H3]2- 375.5[Fe3(Cit)3H]2- 366.5 [Fe3(Cit)4H6]- 926 (H4Cit)n[Fe3O(Cit)3H3]2- 663.5, 759.5(H4Cit)2[Fe3(Cit)3H]2- 558.5, 654.5, 750.5 (H4Cit)[Fe4(Cit)4H]3- 390 [Fe3(Cit)4H5]2- 462.5[Fe3O(Cit)3H3]2- 375.5 [Fe3(Cit)4H6]- 926(H4Cit)4[Fe3O(Cit)3H3]2- 759.5 (H4Cit)[Fe4(Cit)4H]3- 390[Fe3(Cit)4H5]2- 462.5[Fe3(Cit)4H6]- 926(H4Cit)[Fe4(Cit)4H]3- 390

1 : 4 pH 4.3 pH 7.3 pH 9.0[Fe(Cit)2H3]2- 217.5 [Fe(Cit)2H3]2- 217.5 [Fe(Cit)2H3]2- 217.5[Fe(Cit)2H4]- 436 [Fe(Cit)2H4]- 436 [Fe(Cit)2H4]- 436(H4Cit)[Fe2(Cit)2]2- 340 (H4Cit)n[Fe(cit)2H4]- 628, 820 (H4Cit)n[Fe(Cit)2H4]- 628, 820[Fe2(Cit)2H]- 489 [Fe3(Cit)3H]2- 366.5 (H4Cit)n[Fe3(Cit)3H]2- 558.5, 654.5, 750.5(H4Cit)n[Fe2(Cit)2H]- 873, 1064 (H4Cit)n[Fe3(Cit)3H]2- 558.5, 654.5, 750.5, 846 [Fe3O(Cit)3H3]2- 375.5[Fe2(Cit)3H5]- 681 [Fe3O(Cit)3H3]2- 375.5 (H4Cit)n[Fe3O(Cit)3H3]2- 663.5, 759.5[Fe3(Cit)3H]2- 366.5 (H4Cit)n[Fe3O(Cit)3H3]2- 663.5, 759.5 [Fe3(Cit)4H5]2- 462.5(H4Cit)n[Fe3(Cit)3H]2- 558.5, 654.5, 750.5, 846 (H4Cit)4[Fe3O(Cit)3H3]2-·H2O 768.5 [Fe3(Cit)4H6]- 926[Fe3O(Cit)3H3]2- 375.5 [Fe3(Cit)4H5]2- 462.5 (H4Cit)[Fe4(Cit)4H]3- 390[Fe3(Cit)4H5]2- 462.5 [Fe3(Cit)4H6]- 926[Fe3(Cit)4H6]- 926 (H4Cit)[Fe4(Cit)4H]3- 390(H4Cit)[Fe4(Cit)4H]3- 390

1 : 10 pH 4.3 pH 7.3 pH 9.0[Fe(Cit)2H4]- 436 [Fe(Cit)2H3]2- 217.5 [Fe(Cit)2H3]2- 217.5(H4Cit)n[Fe(Cit)2H4]- 628, 820 [Fe(Cit)2H4]- 436 [Fe(Cit)2H4]- 436(H4Cit)[Fe2(Cit)2]2- 340 (H4Cit)n[Fe(Cit)2H4]- 628 (H4Cit)n[Fe(Cit)2H4]- 628, 820(H4Cit)n[Fe2(Cit)2H]- 873, 1064 (H4Cit)n[Fe3(Cit)3H]2- 558.5[Fe2(Cit)3H5]- 681 [Fe3O(Cit)3H3]2- 375.5(H4Cit)n[Fe3(Cit)3H]2- 558.5, 654.5 (H4Cit)3[Fe3O(Cit)3H3]2- 759.5[Fe3O(Cit)3H3]2- 375.5 [Fe3(Cit)4H5]2- 462.5[Fe3(Cit)4H5]2- 462.5 (H4Cit)[Fe4(Cit)4H]3- 390

1 : 20 pH 4.4 pH 7.3 pH 9.0[Fe(Cit)2H4]- 436 [Fe(Cit)2H4]- 436 [Fe(Cit)2H4]- 436(H4Cit)n[Fe(Cit)2H4]- 628, 820 (H4Cit)n[Fe(Cit)2H4]- 628, 820 (H4Cit)n[Fe(Cit)2H4]- 628, 820(H4Cit)[Fe2(Cit)2]2- 340 [Fe3O(Cit)3H3]2- 375.5[Fe2(Cit)3H5]- 681 [Fe3(Cit)4H5]2- 462.5(H4Cit)n[Fe3(Cit)3H]2- 558.5, 654.5, 750.5, 846[Fe3(Cit)4H5]2- 462.5

1 : 100 pH 4.3 pH 7.1 pH 9.0[Fe(Cit)2H4]- 436 [Fe(Cit)2H4]- 436 [Fe(Cit)2H4]- 436(H4Cit)n[Fe(Cit)2H4]- 628, 820 (H4Cit)n[Fe(Cit)2H4]- 628, 820 (H4Cit)n[Fe(Cit)2H4]- 628, 820

This journal is © The Royal Society of Chemistry 2009 Dalton Trans., 2009, 8616–8625 | 8619

at m/z = 340 and 681 may either be attributed to the citricacid adducts (H4Cit)[Fe2(Cit)2]2- and (H4Cit)[Fe2(Cit)2H]- or tothe presence of dinuclear tricitrate complexes [Fe2(Cit)3H4]2- and[Fe2(Cit)3H5]-. Tandem mass spectrometry (MS/MS) studies wereperformed in order to break the possible citric acid adducts andestablish the iron : citrate ratio in the complex; however, these wereinconclusive. Peaks at m/z values 375.5, 462.5, 558.5 and 654.5revealed the presence of a trinuclear species. The mononuclearcomplex [Fe(Cit)2]5- appeared under different protonation statesas an adduct with one or more citric acid molecules at m/z values436, 628 and 820.

When the pH of the solutions was raised to neutral values(Fig. 1(B)) the dominance of the mononuclear species becomesmore evident with peaks at m/z values of 217.5 and 436, andof the respective adducts with citric acid at m/z 628 and 820.Despite the predominance of this species, several trinuclear andone tetranuclear complexes (m/z 390) were also identified in thissolution. It was noticeable that the dinuclear species were absentfrom these spectra, suggesting that these complexes are unlikelyto exist under physiological conditions. The trend suggesting thatmore basic pH values favour the existence of the mononuclearcomplex is confirmed in Fig. 1(C), which corresponds to pH 9.0.A significant increase in the intensity of the peaks correspondingto the mononuclear complex was observed, while most of the mult-inuclear species disappeared. Only the complex [Fe3O(Cit)3H3]2-

was observed as a low intensity peak at m/z value 375.5.At physiological pH values, ferric citrate speciation seems to be

mostly dependent on the excess of citrate available in solution. Ata Fe(III) : citrate molar ratio equal to 1 : 2 (Fig. 2(A)), even thoughthe mononuclear species was still observable, the mass spectrumwas dominated by the presence of oligomeric species, mainlytrinuclear complexes. [Fe3O(Cit)3H3]2- was the most abundantspecies in solution under these conditions appearing at m/z =375.5 and as adducts with citric acid at m/z values 663.5 and 759.5.Increasing the excess of citrate in solution appears to favour theformation of the mononuclear species. At an iron : citrate molarratio of 1 : 10 (Fig 2(B)) the presence of the monoiron dicitratecomplex dominated the spectrum, but trinuclear complexes couldstill be detected. The mononuclear complex was observed inits monocharged form [Fe(Cit)2H4]- (m/z = 436) and as anadduct with one or two citric acids (m/z = 628 and m/z =820 respectively). Fig. 2(C) shows that the dominance of themononuclear species increases with the decreasing iron : citratemolar ratio, with the corresponding disappearance of most of theoligomeric species.

As a result of protonation occurring during the electrosprayionization process, species could be detected with different degreesof protonation. This excludes mass spectrometry as a suitabletechnique for gaining information concerning the ionization stateof the complexes in solution. However, due to the soft ionizationprocess, ESI-MS is a robust method for the establishment of thenuclearity and the Fe(III) : citrate stoichiometry of the differentcomplexes. The isotopic pattern of each detected species was foundto be in accordance with the results of isotopic pattern calculations.

These results are in close agreement with those previouslyreported,23 and clear trends may be deduced from the availabledata. Excess citrate favours the formation of the monoirondicitrate species while with lower citrate concentrations oligomericcomplexes predominate. It is significant that at iron : citrate ratios

Fig. 2 Electrospray ionization-mass spectra of ferric citrate solutionsat pH = 7.4 and different iron : citrate molar ratios. (A) 1 : 2; (B) 1 : 10;(C) 1 : 20. The increasing excess of citrate favours the predominance of themonoiron dicitrate species [Fe(Cit)2H4]-. Spectra are normalized for themost abundant peak. In Fig. (B) and (C) the scale of peak abundance is notshown to 100%, in order to facilitate the visualization of the less abundantspecies.

of 1 : 100 only the [Fe(Cit)2H4]- complex could be detected byESI-MS, regardless of the pH value of the solution under study.Increasing the pH value also favoured the presence of mononucleariron complexes in solution, particularly if the concentration ofcitrate exceeds that of iron by ten times or more.

Overall, the mass spectrometry studies carried out on iron citrateaqueous solutions indicate that at high iron : citrate molar ratiosand low pH values the speciation of these systems is dominatedby oligomeric complexes such as the dinuclear and trinuclearcomplexes (Fig. 3(A) and Fig. 3(B) respectively), competingwith the formation of a mononuclear dicitrate complex, whichtends to dominate at lower iron : citrate molar ratios and higherpH values (Fig. 3(C)). The structure depicted for the dinuclear

8620 | Dalton Trans., 2009, 8616–8625 This journal is © The Royal Society of Chemistry 2009

Fig. 3 Schematic representation of the structure of the ferric citratecomplexes existing in aqueous solution. (A) [Fe2(Cit)2]2-, M2L2; (B)[Fe3(Cit)3]3-, M3L3; (C) [Fe(Cit)3]5-, ML2; (D) [Fe(HCit)]0, MLH.

and mononuclear complexes is a schematic representation ofthe structure obtained by X-ray crystallography.20,22 No crystalstructure is available for the trinuclear species, but its structurewas proposed23 to resemble that of a nonanuclear complex whichcontains a triple layered structure, each layer containing threeiron atoms.21 The provenance of these species as breakdownproducts of the large nonairon complex has been proposed, butthe absence of any parent peak at a m/z value corresponding to acomplex of the type [Fe9(Cit)8Hx](7-x)- strongly suggests that thesespecies are rather likely to be the precursor building blocks of thenonanuclear complex, than to result from its degradation.23 Inthis work, an extensive search for the presence of the nonanuclearcomplex was undertaken and despite the utilization of high massresolution (ESI-FT-ICR-MS) or a different ionisation procedure(MALDI-TOF-MS), its existence remained unconfirmed. Fur-ther, the nonairon complex could not be detected even in solutionsprepared by dissolving crystals of this complex in water. Thedark orange crystals were prepared as described by Bino et al.21

When water was added to the complexes the initial solutionadopted a similar colour, but after complete dissolution of thecrystals (12 h), the solution colour changed to green. This colourchange could be interpreted as a rearrangement of the iron–citratespeciation, suggesting that the nonanuclear complex is only stablein the specific crystallization conditions of low pH values, highiron : citrate molar ratios and high iron concentrations.

Electron paramagnetic resonance spectroscopy

Ferric iron is a d5 element possessing an unpaired electron thatcan be detected utilizing electron paramagnetic resonance spec-troscopy (EPR). Mononuclear ferric complexes typically presenta characteristic signal near g = 4.3 often used to identify their

presence in solution.29 In contrast, when multinuclear complexesoccur, EPR signals usually broaden out or disappear due tospin coupling effects. Considering that ESI-MS results revealedthe mononuclear complex [Fe(Cit)2]5- as the predominant speciesin aqueous solutions at physiological pH values, its abundanceincreasing with the excess of citrate, these mixtures are predictedto possess an EPR signal near g = 4.

With iron at 10 mM (Fig. 4(A)) an EPR signal was foundfor all solutions with an iron–citrate ratio equal to or lowerthan 1 : 10 with the intensity increasing with increasing excess ofcitrate. The signal increase observed when the iron : citrate molarratio was changed from 1 : 100 to 1 : 1000 was probably due tothe conversion of multinuclear species, existing at concentrationsundetectable by ESI-MS, to the mononuclear species. At aniron : citrate ratio of 1 : 5 the EPR signal was only found for thesolution with an iron concentration of 100 mM (Fig. 4(B)). Thehigher concentration mixtures showed EPR signals within theentire range of iron : citrate molar ratios investigated, confirmingthe trend previously described. The signal observed at the higheriron : citrate ratio suggests that its absence in Fig. 4(A) for the1 : 5 condition is likely to result from limits of detection ratherthan the species not being present. This result differs slightlyfrom a previously reported study,30 but differences are most likelydue to the more sensitive instrumentation utilized in the currentinvestigation.

Together, these results agree with those obtained with ESI-MS. At physiological pH values, the mononuclear complex[Fe(Cit)2]5- exists throughout the range of iron : citrate molar ratiosinvestigated, becoming the dominant species for ratios of 1 : 10 orlower, when the EPR signal is well defined. This species accountsfor the dominant peaks observed by mass spectrometry.

Determination of the iron(III) affinity constants of citric acid by thespectrophotometric method

The pKa value for the hydroxyl group of citric acid has beenrecently determined by 13C NMR26 and this new informationwas utilized in the current investigation. For the affinity constantcalculations the ligand (L) was defined as the tetrabasic citrateion ((Cit)4-; C6H4O7

4-), while in previous studies it was alwaysconsidered as the tribasic ion ((Cit)3-; C6H5O7

3-). The pKa valuesfor the three carboxylic acids are pKa1 = 3.13, pKa2 = 4.76 andpKa3 = 6.40 and that for the hydroxyl proton is pKa4 = 14.4.25,26

Spectrophotometric titration was adopted for the stabilityconstant determination and conditions of a high excess of citricacid were utilized, thereby favouring the mononuclear species. Thespectrophotometric data in conjunction with the knowledge of thesolution speciation obtained by mass spectrometry allowed theidentification of the species responsible for the visible spectrumchanges. The species involved are: the partially coordinatedmononuclear complex MLH, multinuclear complexes either M2L2

or M3L3 (there are two models) and the fully coordinated mononu-clear complexes ML2H2, ML2H and ML2. The structures of thesespecies are presented in Fig. 3 with the titration experimentaldata presented in Fig. 5. In Fig. 5(A), the peak shift from358 nm to 338 nm is monitored when the solution is acidifiedand FeCl3 formed. Fig. 5(B) shows the formation of multinuclearcomplexes from the partly coordinated iron(III) citrate complex.In Fig. 5(C) several isosbestic points exist which indicate the

This journal is © The Royal Society of Chemistry 2009 Dalton Trans., 2009, 8616–8625 | 8621

Fig. 4 EPR spectroscopy on iron(III) citrate solutions with varying iron : citrate molar ratios. (A) [Fe3+] = 10 mM; (B) [Fe3+] = 100 mM. The gain settingsfor all spectra have been adjusted to the same concentration of iron. Other conditions of measurement: temperature 12.5 K, microwave power 20 mW,frequency 9.40 GHz, modulation amplitude 0.476 mT, modulation frequency 100 kHz.

transition from multinuclear complexes to the protonated fullycoordinated iron(III) citrate complex (ML2H2) and then to theunprotonated fully coordinated iron(III) citrate complex (ML2).The excess of citric acid utilized during the last titration step(iron : citrate = 1 : 200), together with the MS and EPR evidencestrongly indicates that the mononuclear complex is the singlespecies in solution.

The existence of multiple isosbestic points in the spectraresulting from spectrophotometric titration (Fig. 5(C)) led to theconclusion that in the pH range at which oligomeric complexespredominate, several species coexist. Mass spectrometry datasupports this assumption and indicates that the most abundantspecies should be the dinuclear and trinuclear complexes. Thisinformation led to the adoption of two speciation models forthe determination of the iron(III) citrate stability constants. Thefirst model assumed that M2L2 ([Fe2(Cit)2]2-) is the predominantoligomeric species, while in the second model M3L3 ([Fe3(Cit)3]3-)is considered to dominate. The stability constants determinedby applying both models are presented in Table 2. The differentprotonation states of the ML2 complex are thought to arise fromthe protonation of the free carboxylate groups (Fig. 3(C)) and notfrom the protonation of the citrate hydroxyl function.

With these results, both models were simulated by HySS27 underphysiological conditions corresponding to low and maximumvalues of NTBI; namely 1 mM and 10 mM iron(III) (Fig. 6). Thesespeciation plots demonstrate that at physiological conditions(pH 7.4) when NTBI is 1 mM, the dominant form of iron(III) is themononuclear complex in both models, where as 10 mM iron(III) thecontribution of the multinuclear complexes make an appreciablecontribution in both models.

Discussion

Mass spectrometry is not able to provide information on theprotonation state of the various ferric citrate species existing inaqueous solution, but it can provide evidence for the nuclearity andiron : citrate ratio in the detected complexes. The results obtainedin the course of this work correlate well with the MS speciationreported previously,23 the only significant difference being theinability to find the previously observed dinuclear species at neutralto basic pH values.23 Slight differences in the instrumentationprocedure could enhance or hinder the ionisation of the differentcomplexes, making it possible that less abundant species, as isthe case with dinuclear complexes, might have not been detected.However, both studies are characterized by the predominanceof oligomeric species, particularly trinuclear complexes, at highiron : citrate molar ratios. It is significant, that these trinuclearspecies were identified for the first time by mass spectrometryand have never been crystallized or previously identified by othermethods. However, similar trinuclear complexes with citric acidhave been observed in aqueous solutions with aluminium byNMR.31,32

The mass spectrometry results demonstrate that the mononu-clear dicitrate complex predominates when the molar citrateto Fe(III) ratio is greater than 10, while small oligomericcomplexes increase in abundance at higher iron : citrate molarratios. The solution pH value was found to have a similareffect on the ferric citrate speciation; lower pH values favour-ing the existence of oligomeric species, while increasing thepH value enhances the dominance of the mononuclear ironcomplexes.

8622 | Dalton Trans., 2009, 8616–8625 This journal is © The Royal Society of Chemistry 2009

Fig. 5 Spectrophotometric titration to determine the affinity constantsof ferric citrate species. (A) MLH, [Fe3+] = 460 mM [Citrate] = 8.84 mM,ratio L : M = 19; (B) M2L2 or M3L3, [Fe3+] = 475 mM [Citrate] = 9.12 mM,ratio L : M = 19; (C) ML2, ML2H and ML2H2, [Fe3+] = 435 mM [Citrate] =87.7 mM, ratio L : M = 200.

Concerning the affinity constants for the different complexes,there is a range of published values and speciation models(Table 2). A common source of misunderstanding associated withmost of these studies is the protonation state of the alcoholfunction of citrate upon coordination to Fe(III). The availablecrystal structures for ferric citrate complexes indicate that thehydroxyl proton is lost when citrate coordinates iron(III)20,22 anddefinitive evidence that this group exists in the alkoxide form inaqueous ferric citrate complexes was obtained utilizing high-fieldEPR spectroscopy.33 It is important to note that the high-fieldEPR study was carried out at iron : citrate molar ratios of 1 : 20and under these conditions it was unequivocally demonstratedthat the species present in solution is the mononuclear dicitratecomplex.

The deprotonation of the hydroxyl group of citric acid in ferriccitrate complexes poses a major challenge for the determinationof the stability constants of these species. Most studies treat

citrate as a tribasic ligand (H3L), describing the metal ion induceddeprotonation of the hydroxyl group in the formed complex.However, the determination of this deprotonation at neutralpH values is complicated, limiting the study of iron(III)–citrateaqueous solution to acidic medium.16 The recent determinationof the fourth pKa value of citric acid (14.4)26 renders it possibleto consider H4L in the affinity constants calculations, thusaccounting for the hydroxyl deprotonation.

With the knowledge of the four protonation constants of citricacid and a comprehensive understanding of the complexes presentin solution, the affinity constants for the iron(III) citrate com-plexes were determined utilizing the spectrophotometric titrationmethod. From MS data, it is known that several multinuclearcomplexes exist in solution and no condition has been identifiedunder which one of these species exists individually. This led tothe adoption of two alternative models for data treatment andconstant determination, one stoichiometry includes multinuclearcomplexes where M2L2 largely predominates and the secondconsiders multinuclear complexes based on M3L3. A large excessof citrate was utilized in this study in order to favour the formationof the ML2 species and reduce the number of different complexesexisting in solution.

The values found for the monoiron species MLH, ML2H2,ML2H and ML2, are in good agreement independently of themodel adopted (Table 2). The speciation of ferric citrate multin-uclear complexes is complicated, with several species coexistingin solution, and the two species indicated in Table 2 should justbe regarded as a working model. Due to the multiplicity of ferriccitrate complexes existing under most conditions it is difficult todetermine the affinity constants of the multinuclear complexeswith a high degree of certainty. However, the fact that the differencebetween the affinity constants for the ML2 species as calculated bythe two stoichiometric models is within the respective experimentalerrors indicates that this was a good working hypothesis. This isthe first time that a value for the affinity constant of [Fe(Cit)2]5- hasbeen calculated. The value is particularly important34 consideringthat this complex is likely to have biological relevance in bacteria,plant and animals.

A direct comparison of the values determined in the currentwork with those previously reported is difficult, because in thiswork citrate is defined as H4L as opposed to H3L (most literaturemodels). The values obtained in the course of our investigationare larger than those previously reported, as they have to accountfor the unfavourable energy associated with the deprotonation ofthe hydroxyl groups in citric acid. A useful means of comparingthe different stability constants is the determination of the pFe3+

values, a measure of the aqueous Fe3+ ion existing in the presenceof the ligand under defined conditions. The calculated pFe3+ valuefor citrate at pH = 7.4 under the condition 10 mM citrate with 1 mMFe(III) is 15.0. Based on the literature constants, a value of 16.75has been calculated, but as noted by Harris,34 since the [Fe(Cit)2]5-

could not be included in the calculation, this pFe3+ value shouldonly be considered as an approximation.34

Citrate has been shown to be an NTBI ligand by NMRspectroscopy8 and the understanding of the speciation of ferriccitrate chemistry will help to improve the knowledge relating tothe chemical nature of this iron pool. Considering that citrate ispresent in the serum at an approximate concentration of 100 mM16

and NTBI is typically present at concentrations up to 10 mM35

This journal is © The Royal Society of Chemistry 2009 Dalton Trans., 2009, 8616–8625 | 8623

Table 2 Affinity constants for the Fe(III)–citrate complexes, log bnpra

ML MLH MLH-1 MLH-2 ML2 ML2H ML2H2 ML2H-1 M2L2 M2L2H-2 M3L3 Ref.

24.84 (0.04) 32.73f (0.05) 38.74 (0.05) 43.53 (0.03) 48.0 (0.01) This workb ,c

24.84 (0.04) 32.75f (0.06) 38.77 (0.05) 43.66 (0.02) 73.8 (0.05) This workb ,d

-6.3 10e

11.38 9.46 11e

11.18 12.37 8.49 12e

11.85 11.44 9.40 1.90 13e

11.40 21.17 14e

16.2g ,h 15e

9.5 7.31 15.30g 19.12 10.46g 16e

21.20 17e

a bMpLqHr = [MpLqHr]/[M]p[L]q[H]r. b Affinity constants determined in the course of the reported investigation, citric acid was treated as a tetrabasicligand (H4L) and two speciation models were adopted, standard deviation for the affinity constant determination are indicated within brackets. c Thespeciation model adopted included M2L2 as the multinuclear species. d The speciation model adopted included M3L3 as the multinuclear ferric citratecomplex. e Values found in the literature were determined for citric acid = H3L, the deprotonation of the hydroxyl group has not been considered. f Inthe present study ML2 represents the [Fe(Cit)2]5- complex. g In the literature model ML2 represents the [Fe(HCit)2]3- complex and ML2H-1 stands for[Fe(HCit)2(OH)]4-. h This value was estimated by the author from the equilibrium constant for the aluminium citrate complex.

Fig. 6 Speciation plots for ferric citrate species in aqueous solution when the iron : citrate ratio is 1 : 100 or 1 : 10. Two speciation models were adoptedwhen predicting speciation: the first including the species M3L3 as the predominant oligomeric species existing in solution, while the second includedthe M2L2 complex. (A) M3L3, [Cit] = 100 mM [Fe3+] = 1 mM; (B) M3L3, [Citrate] = 100 mM [Fe3+] = 10 mM; (C) M2L2, [Citrate] = 100 mM [Fe3+] = 1 mM;(D) [Citrate] = 100 mM [Fe3+] = 10 mM.

two speciation models were simulated by HySS for the ironconcentrations of 1 and 10 mM and the physiological concentrationof citrate (Fig. 6). These speciation plots demonstrate that whenthe iron concentration is 1 mM (Fig. 6(A) and Fig. 6(C)) atphysiological pH values (7.4) the predominant species is themonoiron dicitrate species regardless of the adopted model.

Under such conditions (iron : citrate molar ratio of 1 : 100), thecalculated pFe3+ value is 16.2, a value sufficiently high to ensurethat citrate is able to solubilize ferric iron at pH 7.4. When theiron concentration is raised to 10 mM (Fig. 6(B) and Fig. 6(D))it is predicted that multinuclear complexes coexist with themonoiron species, as demonstrated by mass spectrometry for

8624 | Dalton Trans., 2009, 8616–8625 This journal is © The Royal Society of Chemistry 2009

the same iron : citrate molar ratios. The speciation plots are ingood agreement with the available MS data for the correspondingiron : citrate molar ratios. It is important to note that regardingmultinuclear species, the relative abundances represented in thespeciation plots correspond to the abundance of a pool of differentcomplexes and not the concentration of individual species. Massspectrometry demonstrates that the multinuclear species constitutea heterogeneous pool (Table 1) which is represented as a singlespecies in these speciation plots.

Conclusion

The data presented in this communication demonstrate that thepredominant ferric citrate species at physiological pH values arethe monoiron dicitrate complex and multinuclear species of lownuclearity, in particular trinuclear complexes. The [Fe(Cit)2]5-

complex predominates in the iron : citrate molar ratio range 1 : 100to 1 : 10, suggesting that this is the most relevant small molecularweight NTBI species. The presence of oligomeric species becomesappreciable when iron : citrate ratios are higher than 1 : 10 (NBTI >

10 mM).

Acknowledgements

Authors would like to thank Dr Chris Kay from University CollegeLondon for providing EPR facilities. Silva A. M. N. would liketo acknowledge “Fundacao para a Ciencia e Tecnologia”-FCT,Lisboa, Portugal for his PhD grant [SFRH/BD/22633/2005].

Notes and references

1 J. J. Villafranca and A. S. Mildvan, J. Biol. Chem., 1972, 247,3454–3463.

2 D. L. Bakkeren, C. M. H. De Jeu-Jaspers, C. Van Der Heul and H. G.Van Eijk, Int. J. Biochem., 1985, 17, 925–930.

3 L. O. Tiffin, Plant Physiol., 1966, 41, 51–54.4 J. F. Briat, C. Curie and F. Gaymard, Curr. Opin. Plant Biol., 2007, 10,

276–282.5 A. D. Ferguson, R. Chakraborty, B. S. Smith, L. Esser, D. Van Der

Helm and J. Deisenhofer, Science, 2002, 295, 1715–1719.6 M. L. Guerinot, E. Meidl and L. Plessner, J. Bacteriol., 1990, 172,

3298–3303.7 H. Drechsel and G. Winklemann, in Transition Metals in Microbial

Metabolism, ed. G. Winklemann and C. Carrano, Harwood AcademicPublishers, 1st edn, 2000.

8 M. Grootveld, J. D. Bell, B. Halliwell, O. I. Aruoma, A. Bomford andP. J. Sadler, J. Biol. Chem., 1989, 15, 4417–4422.

9 J. L. Pierre and I. Gautier-Luneau, BioMetals, 2000, 13, 91–96.10 O. E. Lanford and J. R. Quinan, J. Am. Chem. Soc., 1948, 70, 2900–

2903.11 R. C. Warner and I. Weber, J. Am. Chem. Soc., 1953, 75, 5086–5094.12 T. B. Field, J. L. McCourt and W. A. E. McBride, Can. J. Chem., 1974,

52, 3119–3124.13 R. E. Hamm, C. M. Shull Jr. and D. M. Grant, J. Am. Chem. Soc.,

1954, 76, 2111–2114.14 C. F. Timberlake, J. Chem. Soc., 1964, 5078–5085.15 R. B. Martin, J. Inorg. Biochem., 1986, 28, 181–187.16 L. C. Konigsberg, E. Konigsberg, P. May and G. T. Hefter, J. Inorg.

Biochem., 2000, 78, 175–184.17 Y. Z. Hamada, N. Bayakly, A. Peipho and B. Carlson, Synth. React.

Inorg., Met.-Org., Nano-Met. Chem., 2006, 36, 469–476.18 T. G. Spiro, L. Pape and P. Saltman, J. Am. Chem. Soc., 1967, 89,

5555–5558.19 T. G. Spiro, L. Pape and P. Saltman, J. Am. Chem. Soc., 1967, 89,

5559–5561.20 I. Shweky, A. Bino, D. P. Goldberg and S. J. Lippard, Inorg. Chem.,

1994, 33, 5161–5162.21 A. Bino, I. Shweky, S. Cohen, E. R. Bauminger and S. J. Lippard, Inorg.

Chem., 1998, 37, 5168–5172.22 M. Matzapetakis, C. P. Raptopoulou, A. Tsohos, V. Papaefthymiou, N.

Moon and A. Salifoglou, J. Am. Chem. Soc., 1998, 120, 13266–13267.23 I. Gautier-Luneau, C. Merle, D. Phanon, C. Lebrun, F. Biaso, G.

Serratrice and J. L. Pierre, Chem.–Eur. J., 2005, 11, 2207–2219.24 P. Gans and A. Sabatini, Ann. Chim., 1999, 89, 45–49.25 A. E. Martell and R. M. Smith, Critical stability constants V.3, Plenum

Press, New York, 1977.26 A. M. N. Silva, X. L. Kong and R. C. Hider, BioMetals, 2009,

DOI: 10.1007/s10534-009-9224-5.27 L. Alderighi, P. Gans, A. Ienco, D. Peters, A. Sabatini and A. Vacca,

Coord. Chem. Rev., 1999, 184, 311–318.28 C. F. Baes and R. E. Mesmer, in The hydrolysis of cations, Wiley, New

York, 1st edn, 1976.29 W. R. Hagen, Biomolecular EPR Spectroscopy, CRC Press, Boca Raton,

FL, USA, 1st edn, 2009.30 R. W. Evans, R. Rafique, A. Zarea, C. Rapisarda, R. Cammack, P. J.

Evans, J. B. Porter and R. C. Hider, JBIC, J. Biol. Inorg. Chem., 2007,13, 57–74.

31 A. Lakatos, I. Banyai, P. Decock and T. Kiss, Eur. J. Inorg. Chem.,2001, 461–469.

32 M. Matzapetakis, M. Kourgiantakis, M. Dakanali, C. P. Raptopoulou,A. Terzis, A. Lakatos, T. Kiss, I. Banyai, L. Iordanidis, T. Mavromous-takos and A. Salifoglou, Inorg. Chem., 2001, 40, 1734–1744.

33 F. Biaso, C. Duboc, B. Barbara, G. Serratrice, F. Thomas, D. Charapoffand C. Beguin, Eur. J. Inorg. Chem., 2005, 467–478.

34 W. R. Harris, in Molecular and cellular iron transport, ed. D. M. Tem-pleton, Marcel Dekker Inc., New York, 2002.

35 W. Breuer, C. Hershko and Z. I. Cabantchik, Transfus. Sci., 2000, 23,185–192.

This journal is © The Royal Society of Chemistry 2009 Dalton Trans., 2009, 8616–8625 | 8625